[Geopackage] Preparing for GeoPackage 1.0.2

Brad Hards bradh at frogmouth.net
Sat Dec 26 20:25:24 EST 2015


On Sat, 26 Dec 2015 11:21:04 AM Jeff Yutzler wrote:
> All,
> The SWG has recently voted to recommend to the OGC Technical Committee to
> adopt the working version of the GeoPackage specification as Version 1.0.2.
> For more details, please see my blog post:
> http://geopackage.blogspot.com/2015/12/preparing-for-geopackage-102.html
There are two issues that you aren't considering in that text.

The first is that the change in #132 means that any implementation that has 
extensions will be generating non-compliant gpkg files (since "The definition 
column value in a gpkg_extensions row for those extensions SHALL contain the 
sub-annex name and as a reference "). That isn't likely to cause problems (but 
then, it probably could be "should") but it would be worth noting.

The other is that #137 is a breaking extension, which isn't what extensions 
should be for (extend, not change). There is a warning (even in bold!) but 
there is no way that the producer can possibly check that all consumers know 
how to handle new-style CRS WKT. It might have been possible in a network 
scenario, with some kind of protocol negotiation, but it doesn't work in a 
mass-market offline tool.

I do have a suggestion, and that is that the extension should add a new column 
to the gpkg_spatial_ref_sys table (srtext2, or whatever name the committee 
would like to agree on), so that consumers that can handle the new-style WKT 
if available, but would still be able to use the old WKT if they haven't been 
updated.

I know you want to get 1.0.2 out, but I think a little more work on that 
extension is warranted.

Brad


More information about the Geopackage mailing list